Event Grid inference workflow returned intermittent HTTP 500

Microsoft cannot perform an account-specific Event Grid backend-telemetry lookup through the public Q&A forum; open an Azure technical support request against the affected Event Grid domain and provide the UTC timestamp and x-ms-request-id.

Open the investigation request

1. In the Azure portal, open the affected Event Grid domain.
2. Select Support + Troubleshooting under Help.
3. If the suggested diagnostics do not resolve it, select Create a support request.
4. Choose:
   • Issue type: Technical
   • Service: Azure Event Grid
   • Resource: The affected domain
   • Problem type: Use the closest available category, such as Publisher issues, Subscriber issues, or Development issues
   • Severity: Match the production impact

Creating a technical request requires an applicable support plan and subscription-level Owner, Contributor, Support Request Contributor, or a custom role containing Microsoft.Support/* permissions.^1^

Include the following in Problem details:

• Full Event Grid domain resource ID and endpoint hostname
• Exact failure timestamp in UTC, including seconds and the relevant time window
• Complete x-ms-request-id
• HTTP status and complete response headers/body, with secrets removed
• Whether the 500 occurred while publishing to Event Grid, during Event Grid delivery, or in the inference endpoint
• Event subscription name and destination resource ID, if this was a delivery failure
• Region, recurrence pattern, retry outcomes, and business impact
• Any application trace/correlation ID spanning Event Grid and the inference service

Ask explicitly:

Please correlate this x-ms-request-id and UTC timestamp with Event Grid service telemetry and identify, if retained and releasable, the regional backend cluster/deployment stamp that processed the request and whether the HTTP 500 originated in Event Grid or the downstream endpoint.

Do not post subscription IDs, resource IDs, request traces, tokens, event payloads, or other account-specific data publicly. Submit them through the support case.

Grant diagnostic access

In the request’s Advanced diagnostic information section, select Yes. Azure Support can access advanced diagnostics only when the case is created through the Azure portal and permission is granted.^2^

If the case is already open:

1. Go to Help + Support → All support requests.
2. Open the request.
3. Select Advanced diagnostic information.
4. Select Yes, then Submit.

The supplied identifiers are suitable correlation evidence, but public documentation does not guarantee that a particular internal cluster or deployment-stamp identifier will be retained or disclosed. Azure Support must determine that from the available service telemetry.

Preserve customer-visible evidence

For future occurrences, enable Event Grid resource logs:

1. Open the Event Grid domain.
2. Go to Monitoring → Diagnostic settings.
3. Select Add diagnostic setting.
4. Select allLogs.
5. Send the logs to a Log Analytics workspace, storage account, or Event Hub.
6. Select Save.

For Event Grid domains, the available categories include DeliveryFailures, PublishFailures, and DataPlaneRequests.^3^ Event Grid also recommends using diagnostic logs and metrics when investigating publish or delivery failures.^4^

These resource logs can establish which stage failed and provide evidence for support, but they are not a public substitute for Microsoft’s internal routing telemetry.

References

1. Create an Azure support request
2. Manage an Azure support request
3. Enable diagnostic logs for Event Grid resources
4. Troubleshoot Azure Event Grid issues